90 lines · plain
1; NOTE: Assertions have been autogenerated by utils/update_analyze_test_checks.py UTC_ARGS: --filter "LV: Found an estimated cost of [0-9]+ for VF [0-9]+ For instruction:\s*store i32 %v4, ptr %out4"2; RUN: opt -passes=loop-vectorize -vectorizer-maximize-bandwidth -S -mattr=+sse2 --debug-only=loop-vectorize --disable-output < %s 2>&1 | FileCheck %s --check-prefix=SSE23; RUN: opt -passes=loop-vectorize -vectorizer-maximize-bandwidth -S -mattr=+avx --debug-only=loop-vectorize --disable-output < %s 2>&1 | FileCheck %s --check-prefix=AVX14; RUN: opt -passes=loop-vectorize -vectorizer-maximize-bandwidth -S -mattr=+avx2 --debug-only=loop-vectorize --disable-output < %s 2>&1 | FileCheck %s --check-prefix=AVX25; RUN: opt -passes=loop-vectorize -vectorizer-maximize-bandwidth -S -mattr=+avx512vl --debug-only=loop-vectorize --disable-output < %s 2>&1 | FileCheck %s --check-prefix=AVX5126; REQUIRES: asserts7 8target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128"9target triple = "x86_64-unknown-linux-gnu"10 11@A = global [1024 x i8] zeroinitializer, align 12812@B = global [1024 x i32] zeroinitializer, align 12813 14define void @test() {15; SSE2-LABEL: 'test'16; SSE2: LV: Found an estimated cost of 1 for VF 1 For instruction: store i32 %v4, ptr %out4, align 417; SSE2: LV: Found an estimated cost of 1 for VF 1 For instruction: store i32 %v4, ptr %out4, align 418; SSE2: LV: Found an estimated cost of 40 for VF 2 For instruction: store i32 %v4, ptr %out4, align 419; SSE2: LV: Found an estimated cost of 84 for VF 4 For instruction: store i32 %v4, ptr %out4, align 420; SSE2: LV: Found an estimated cost of 168 for VF 8 For instruction: store i32 %v4, ptr %out4, align 421;22; AVX1-LABEL: 'test'23; AVX1: LV: Found an estimated cost of 1 for VF 1 For instruction: store i32 %v4, ptr %out4, align 424; AVX1: LV: Found an estimated cost of 1 for VF 1 For instruction: store i32 %v4, ptr %out4, align 425; AVX1: LV: Found an estimated cost of 24 for VF 2 For instruction: store i32 %v4, ptr %out4, align 426; AVX1: LV: Found an estimated cost of 46 for VF 4 For instruction: store i32 %v4, ptr %out4, align 427; AVX1: LV: Found an estimated cost of 95 for VF 8 For instruction: store i32 %v4, ptr %out4, align 428; AVX1: LV: Found an estimated cost of 190 for VF 16 For instruction: store i32 %v4, ptr %out4, align 429;30; AVX2-LABEL: 'test'31; AVX2: LV: Found an estimated cost of 1 for VF 1 For instruction: store i32 %v4, ptr %out4, align 432; AVX2: LV: Found an estimated cost of 1 for VF 1 For instruction: store i32 %v4, ptr %out4, align 433; AVX2: LV: Found an estimated cost of 24 for VF 2 For instruction: store i32 %v4, ptr %out4, align 434; AVX2: LV: Found an estimated cost of 46 for VF 4 For instruction: store i32 %v4, ptr %out4, align 435; AVX2: LV: Found an estimated cost of 95 for VF 8 For instruction: store i32 %v4, ptr %out4, align 436; AVX2: LV: Found an estimated cost of 190 for VF 16 For instruction: store i32 %v4, ptr %out4, align 437;38; AVX512-LABEL: 'test'39; AVX512: LV: Found an estimated cost of 1 for VF 1 For instruction: store i32 %v4, ptr %out4, align 440; AVX512: LV: Found an estimated cost of 1 for VF 1 For instruction: store i32 %v4, ptr %out4, align 441; AVX512: LV: Found an estimated cost of 7 for VF 2 For instruction: store i32 %v4, ptr %out4, align 442; AVX512: LV: Found an estimated cost of 14 for VF 4 For instruction: store i32 %v4, ptr %out4, align 443; AVX512: LV: Found an estimated cost of 21 for VF 8 For instruction: store i32 %v4, ptr %out4, align 444; AVX512: LV: Found an estimated cost of 35 for VF 16 For instruction: store i32 %v4, ptr %out4, align 445; AVX512: LV: Found an estimated cost of 70 for VF 32 For instruction: store i32 %v4, ptr %out4, align 446; AVX512: LV: Found an estimated cost of 140 for VF 64 For instruction: store i32 %v4, ptr %out4, align 447;48entry:49 br label %for.body50 51for.body:52 %iv = phi i64 [ 0, %entry ], [ %iv.next, %for.body ]53 54 %iv.0 = add nuw nsw i64 %iv, 055 %iv.1 = add nuw nsw i64 %iv, 156 %iv.2 = add nuw nsw i64 %iv, 257 %iv.3 = add nuw nsw i64 %iv, 358 %iv.4 = add nuw nsw i64 %iv, 459 60 %in = getelementptr inbounds [1024 x i8], ptr @A, i64 0, i64 %iv.061 %v.narrow = load i8, ptr %in62 63 %v = zext i8 %v.narrow to i3264 65 %v0 = add i32 %v, 066 %v1 = add i32 %v, 167 %v2 = add i32 %v, 268 %v3 = add i32 %v, 369 %v4 = add i32 %v, 470 71 %out0 = getelementptr inbounds [1024 x i32], ptr @B, i64 0, i64 %iv.072 %out1 = getelementptr inbounds [1024 x i32], ptr @B, i64 0, i64 %iv.173 %out2 = getelementptr inbounds [1024 x i32], ptr @B, i64 0, i64 %iv.274 %out3 = getelementptr inbounds [1024 x i32], ptr @B, i64 0, i64 %iv.375 %out4 = getelementptr inbounds [1024 x i32], ptr @B, i64 0, i64 %iv.476 77 store i32 %v0, ptr %out078 store i32 %v1, ptr %out179 store i32 %v2, ptr %out280 store i32 %v3, ptr %out381 store i32 %v4, ptr %out482 83 %iv.next = add nuw nsw i64 %iv.0, 584 %cmp = icmp ult i64 %iv.next, 102485 br i1 %cmp, label %for.body, label %for.cond.cleanup86 87for.cond.cleanup:88 ret void89}90